[ADT] Simplify ResolveUnderlyingType (NFC)

[kazutakahirata]

We have three implementations of ResolveUnderlyingType:

• enum
• bool
• neither

This patch combines the latter two with std::conditional_t.

Without this patch, we use "void" to trigger a compilation failure
downstream when sizeof(bool) != 1, which is not very friendly. This
patch instead uses static_assert to catch the case where the user
chooses to use bool but sizeof(bool) != 1.

[llvmbot]

@llvm/pr-subscribers-llvm-adt

Author: Kazu Hirata (kazutakahirata)

Changes

We have three implementations of ResolveUnderlyingType:

• enum
• bool
• neither

This patch combines the latter two with std::conditional_t.

Without this patch, we use "void" to trigger a compilation failure
downstream when sizeof(bool) != 1, which is not very friendly. This
patch instead uses static_assert to catch the case where the user
chooses to use bool but sizeof(bool) != 1.

---

Full diff: https://github.com/llvm/llvm-project/pull/164114.diff

1 Files Affected:

• (modified) llvm/include/llvm/ADT/Bitfields.h (+3-6)

diff --git a/llvm/include/llvm/ADT/Bitfields.h b/llvm/include/llvm/ADT/Bitfields.h
index 1af276145ba48..1fbc41c472581 100644
--- a/llvm/include/llvm/ADT/Bitfields.h
+++ b/llvm/include/llvm/ADT/Bitfields.h
@@ -154,12 +154,9 @@ struct ResolveUnderlyingType {
   using type = std::underlying_type_t<T>;
 };
 template <typename T> struct ResolveUnderlyingType<T, false> {
-  using type = T;
-};
-template <> struct ResolveUnderlyingType<bool, false> {
-  /// In case sizeof(bool) != 1, replace `void` by an additionnal
-  /// std::conditional.
-  using type = std::conditional_t<sizeof(bool) == 1, uint8_t, void>;
+  static_assert(!std::is_same_v<T, bool> || sizeof(bool) == 1,
+                "T being bool requires sizeof(bool) == 1.");
+  using type = std::conditional_t<std::is_same_v<T, bool>, uint8_t, T>;
 };
 
 } // namespace bitfields_details

[llvm-ci]

LLVM Buildbot has detected a new failure on builder clang-m68k-linux-cross running on suse-gary-m68k-cross while building llvm at step 5 "ninja check 1".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/27/builds/17710

Here is the relevant piece of the build log for the reference

Step 5 (ninja check 1) failure: stage 1 checked (failure)
******************** TEST 'LLVM-Unit :: CAS/./CASTests/2/17' FAILED ********************
Script(shard):
--
GTEST_OUTPUT=json:/var/lib/buildbot/workers/suse-gary-m68k-cross/clang-m68k-linux-cross/stage1/unittests/CAS/./CASTests-LLVM-Unit-3204612-2-17.json GTEST_SHUFFLE=0 GTEST_TOTAL_SHARDS=17 GTEST_SHARD_INDEX=2 /var/lib/buildbot/workers/suse-gary-m68k-cross/clang-m68k-linux-cross/stage1/unittests/CAS/./CASTests
--

Script:
--
/var/lib/buildbot/workers/suse-gary-m68k-cross/clang-m68k-linux-cross/stage1/unittests/CAS/./CASTests --gtest_filter=CASProgramTest.MappedFileRegionArenaTest
--
Note: Google Test filter = CASProgramTest.MappedFileRegionArenaTest
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from CASProgramTest
/var/lib/buildbot/workers/suse-gary-m68k-cross/clang-m68k-linux-cross/llvm/llvm/unittests/CAS/ProgramTest.cpp:147: Failure
Value of: PI.ReturnCode == 0
  Actual: false
Expected: true


/var/lib/buildbot/workers/suse-gary-m68k-cross/clang-m68k-linux-cross/llvm/llvm/unittests/CAS/ProgramTest.cpp:147
Value of: PI.ReturnCode == 0
  Actual: false
Expected: true



********************